Optimized Navigation for SLAM Using Marker-Assisted Region Scanning, Path Finding, and Mapping Completion Control

Abstract

This paper introduces Marker-Assisted Region Scanning for Simultaneous Localization and Mapping (MARS-SLAM), a novel approach to optimizing SLAM in unknown environments. Designed to enhance autonomous exploration in extreme conditions, MARS-SLAM ensures efficient navigation while providing a systematic method for verifying mapping completion. The approach leverages virtual markers to track unexplored regions, guiding the robot through an organized and comprehensive exploration process. Markers are placed at the LiDAR sensor’s range limit in free areas, maintaining a dynamic list of regions yet to be visited. Mapping is considered complete when no markers remain, signifying full coverage of the environment. Target marker selection is based on age (creation order) and distance (path length from the robot). The method was validated in three virtual environments of varying complexity, demonstrating superior performance compared to alternative navigation strategies, including predefined zigzag routes and routes generated by Ant Colony Optimization (ACO). Experimental results show that MARS-SLAM achieves complete and accurate mapping while significantly reducing the number of poses required. Specifically, it achieves a 64.39% reduction in poses compared to ACO and 71.07% compared to zigzag navigation, highlighting its efficiency in complex environments.